Skip to content

Harness Engineer Roadmap

A path into the layer that turns a model into an agent — the loop, the system prompt, the tool surface, sandboxing, delegation and memory, built and operated in production.

Entry level
$101,541
AI Engineer Entry Level salary estimate (average, US) — proxy: «Harness Engineer» is not yet a reported job title, and the work is hired under AI Engineer and Agent Infrastructure titles
Mid level
$147,289
AI Agent Engineer salary estimate (average, US) — proxy: measures AI Agent Engineer, the closest reported title to this role
Senior level
$164,244
Senior AI Agent Engineer salary estimate (average, US) — proxy: measures Senior AI Agent Engineer, not a harness-specific title
Glassdoor

A harness is the layer between a model and the world. It decides what the model is told about itself, which tools it can ask for, what happens when it asks for something irreversible, what it remembers at turn sixty, and when it stops. The model supplies judgement. The harness supplies everything else — and it is ordinary code, written and reviewed like any other part of a system.

This path builds one from the loop upward. It starts with the fifteen lines underneath every agent, then treats the system prompt and the tool schemas as versioned artefacts rather than as text boxes, then adds the sandbox and permission layers that make an agent safe to point at a real repository. The last two phases are about scale and operation: delegating to subagents for context isolation, compacting a session that has outgrown its window, and producing traces that say which layer failed when a run goes wrong.

The reason this material is scarce is that harness code is expensive to develop, cheap to copy and invisible in the finished product, so the capable ones stay private. The compensation is that the skill is portable: models change, and the questions of blast radius, tool surface and what survives compaction do not.

New to Linux and the command line?

This path assumes fundamentals you may not have yet. Our Foundations Pack is out and free — Linux, the shell and Git, with exercises that mark your work and explain why you got it wrong. We're writing an agents pack next; leave your email if you want to hear when it ships.

One email when the pack launches. No spam, unsubscribe any time.

Your progress0%

The path, phase by phase

  1. The Agentic Loop

    The mechanism underneath every agent, seen plainly. Done when you can write the loop from memory, branch on the stop reason rather than on the response text, and explain why turn forty costs more than turn one. Skip this and you will debug an agent by editing prose, because the loop will look like a black box rather than fifteen lines you own.

    3-4 weeks
    6 Skills
    Agentic Loop Construction
    Stop Reason Handling
    Message History Management
    Token Accounting
    Anthropic SDK
    Iteration & Spend Bounding
    Show details, projects and resources

    Skills you'll master

    Agentic Loop Constructionintermediate
    Stop Reason Handlingintermediate
    Message History Managementintermediate
    Token Accountingintermediate
    Anthropic SDKintermediate
    Iteration & Spend Boundingintermediate

    Hands-on projects

    1. 01Write an agent loop in under fifty lines that runs a single tool and terminates correctly on every stop reason
    2. 02Bound a loop by both iteration count and cumulative spend, and prove it trips before the budget does
    3. 03Instrument the loop to print input tokens per turn, and chart how they grow across a twenty-turn session
    4. 04Deliberately return a malformed tool result and make the loop recover instead of dying mid-task
  2. System Prompt & Tool Surface

    The two artefacts the model actually programs against. Done when your system prompt lives in version control with a recorded reason for every change, and your tool schemas refuse the calls you do not want instead of asking the model politely not to make them. Teams that treat these as a text box rediscover the same bugs every quarter.

    5-7 weeks
    7 Skills
    System Prompt Design
    Prompt Versioning & Review
    Tool Description Writing
    JSON Schema Design
    Error Messages as Model Input
    Prompt Caching
    Tool Surface Sizing
    Show details, projects and resources

    Skills you'll master

    System Prompt Designadvanced
    Prompt Versioning & Reviewintermediate
    Tool Description Writingadvanced
    JSON Schema Designadvanced
    Error Messages as Model Inputintermediate
    Prompt Cachingintermediate
    Tool Surface Sizingadvanced

    Hands-on projects

    1. 01Put a system prompt under version control and write the commit message that justifies each change for a month
    2. 02Take a failing instruction and move the fix down into a schema constraint that makes the bad call impossible
    3. 03Rewrite a tool's error returns so a failed call steers the next attempt instead of ending the task
    4. 04Measure the token cost of your tool definitions per turn, then cut the surface in half and compare selection accuracy
    5. 05Build a harness whose stable prefix lets caching pay, and measure the difference across a long session
  3. Sandbox, Permissions & Hooks

    The layer that decides what happens when the agent asks for something dangerous. Done when you can state the blast radius of your worst single tool call, confine a path without being fooled by a symlink, and explain why consent fatigue makes a naive approval prompt less safe than no prompt at all. This phase is where an impressive demo becomes something you would point at a real repository.

    5-7 weeks
    7 Skills
    Sandboxing & Process Isolation
    Path Confinement
    Permission Model Design
    Prompt Injection Defence
    Lifecycle Hooks
    Containers
    Least Privilege
    Show details, projects and resources

    Skills you'll master

    Sandboxing & Process Isolationadvanced
    Path Confinementadvanced
    Permission Model Designadvanced
    Prompt Injection Defenceadvanced
    Lifecycle Hooksintermediate
    Containersintermediate
    Least Privilegeadvanced

    Hands-on projects

    1. 01Write a path confinement check that resolves symlinks and survives an adversarial test suite you wrote to break it
    2. 02Classify every tool in your harness by reversibility and wire the irreversible ones to an explicit confirmation
    3. 03Run the agent inside a container with no network egress and get a real task done anyway
    4. 04Plant a prompt injection in a file the agent will read, then show that your permission layer bounds what it achieves
    5. 05Add a post-edit hook that runs the formatter and type-checker, and feed the failures back to the agent as input
  4. Delegation & Long Sessions

    What a harness does when one window is not enough. Done when you delegate for context isolation rather than for the appearance of parallelism, and when a sixty-turn session still knows the task it was given in turn one. Most agents that impress for ten minutes and fall apart over an afternoon fail exactly here.

    4-6 weeks
    6 Skills
    Subagent Architecture
    Context Isolation
    Task Handoff Design
    History Compaction
    Durable Memory
    Concurrency Control
    Show details, projects and resources

    Skills you'll master

    Subagent Architectureadvanced
    Context Isolationadvanced
    Task Handoff Designadvanced
    History Compactionadvanced
    Durable Memoryintermediate
    Concurrency Controlintermediate

    Hands-on projects

    1. 01Delegate a broad search to a subagent and measure the parent's window with and without the delegation
    2. 02Write a handoff prompt precise enough that the subagent needs no follow-up question
    3. 03Implement compaction that preserves the goal, the decisions and their reasons, and drops what can be re-read
    4. 04Run a session past the point where compaction triggers and verify the agent still knows what it was asked
    5. 05Make two subagents work concurrently without editing the same files, and detect it when they try
  5. Operating a Harness in Production

    Everything that separates a harness that works on your machine from one other people depend on. Done when a failed run leaves you a trace that says which layer failed, and when a prompt change ships only after an evaluation says it helped. Without this phase you will be tuning a system whose behaviour you can only describe anecdotally.

    5-7 weeks
    6 Skills
    Agent Tracing & Observability
    Evaluation Harness Design
    Cost & Latency Budgeting
    Regression Testing for Agents
    Failure Triage by Layer
    OpenTelemetry
    Show details, projects and resources

    Skills you'll master

    Agent Tracing & Observabilityadvanced
    Evaluation Harness Designadvanced
    Cost & Latency Budgetingadvanced
    Regression Testing for Agentsadvanced
    Failure Triage by Layeradvanced
    OpenTelemetryintermediate

    Hands-on projects

    1. 01Emit a trace per run that records every tool call, its arguments, its result size and its latency
    2. 02Build a twenty-case evaluation set and gate every system prompt change on it
    3. 03Triage ten real failures and attribute each to the loop, the prompt, a tool, the sandbox or the model
    4. 04Track cost and latency per completed task, not per call, and find the tool that dominates both
    5. 05Assess an agent you did not write from the outside and document what you inferred about its harness

Frequently asked questions

Related roadmaps